How much does an AI API cost in Rupiah?
Cost is computed per request from token rates, not per subscription. This page breaks down the rates, cache rates, worked examples, and payment paths from Indonesia.
Every model uses the same unit: dollar credit per 1 million tokens, deducted from the plan credit as each request completes. Repeated input automatically gets the far cheaper cache rate, and the hard budget stops exactly at the limit.
Token rates per 1 million
Active Rins catalog rates per 1 million tokens. Cache is input that was already processed for a similar request and is billed far cheaper automatically.
| Model | Input | Cache | Output | Cache vs input | Plans from |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flashdeepseek-v4.1-flash-1017 | $0.12 | $0.0024 | $0.48 | 50× cheaper | Rp5,000 |
| GLM-5.3 Flashglm-5.3-flash | $0.12 | $0.0240 | $0.40 | 5× cheaper | Rp5,000 |
| GLM-5.3glm-5.3 | $1.12 | $0.2080 | $3.52 | 5× cheaper | Rp5,000 |
| Kimi K3kimi-k3 | $3.00 | $0.3000 | $15.00 | 10× cheaper | Rp5,000 |
Example: the same 100,000-token prompt resent on DeepSeek costs $0.012 as new input, or $0.00024 as cache; a 50x difference.
How many tokens is $12 of credit?
Daily plan example (Rp5,000, $12 credit) for each model, computed from the rates above and rounded down.
| Model | New input | Cache | Output |
|---|---|---|---|
| DeepSeek V4.1 Flash | 100 million | 5 billion | 25 million |
| GLM-5.3 Flash | 100 million | 500 million | 30 million |
| GLM-5.3 | 10.7 million | 57.6 million | 3.4 million |
| Kimi K3 | 4 million | 40 million | 800 thousand |
The weekly plan ($84 credit) multiplies these numbers by 7; the monthly plan ($100 credit) by about 8.3. Mixed input and output split the credit by each portion at the rates above.
Worked examples
All examples use DeepSeek V4.1 Flash rates: input $0.12, cache $0.0024, output $0.48 per 1 million tokens. The math is written out so you can check it yourself.
| Usage pattern | Calculation | Total credit |
|---|---|---|
| Short chat50 requests | Input 200,000 tokens × $0.12 + output 30,000 tokens × $0.48 | ± $0,04 |
| Coding agent200 requests, 90% cached | New input 800,000 × $0.12 + cache 7,200,000 × $0.0024 + output 400,000 × $0.48 | ± $0,31 |
| Heavy production25 million input tokens per minute, 95% cache, 2 hours | New input 1,250,000 × $0.12 + cache 23,750,000 × $0.0024 per minute, then × 120 minutes | ± $25 per 2 hours |
The heavy-production pattern is a rough upper bound; a $100 monthly plan credit equals roughly eight two-hour sessions at that pattern.
Compare cost paths
The table lists what to check on each path before buying; provider conditions can change.
| Path | Currency and exchange | Payment | Spend limit |
|---|---|---|---|
| Pay the global model provider directly | USD billing; exchange rates and international card fees are on you | Usually an international credit card and a minimum top-up | Depends on the provider panel |
| Global aggregator platform | USD billing, per token | Usually an international card or crypto | Depends on the platform |
| Rins AI Gateway | Final Rupiah pricing, shown before checkout | Local bank transfer or USDT, no international card | Hard budget stops exactly at the limit, with no debt |
Paying from Indonesia
Final Rupiah plan pricing with dollar credit is on the plans page.
Local bank transfer or USDT at checkout, with no international credit card. Methods and fees are shown before confirmation.
Your key is ready within 24 hours of confirmed payment; reveal it again anytime in the dashboard and use it immediately.
Notes
Credit exhausted means stopped. No debt and no automatic subscription; the hard budget stops exactly at the limit.
More details in FAQ, terms of service, and service status.
All rates and examples on this page use the active catalog; prices can change for new orders and are always shown before checkout.